On the challenges of detecting MCI using EEG in the wild
Authors:
Aayush Mishra,
David Joffe,
Sankara Surendra Telidevara,
David S Oakley,
Anqi Liu
Abstract:
Recent studies have shown promising results in the detection of Mild Cognitive Impairment (MCI) using easily accessible Electroencephalogram (EEG) data which would help administer early and effective treatment for dementia patients. However, the reliability and practicality of such systems remains unclear. In this work, we investigate the potential limitations and challenges in developing a robust…
▽ More
Recent studies have shown promising results in the detection of Mild Cognitive Impairment (MCI) using easily accessible Electroencephalogram (EEG) data which would help administer early and effective treatment for dementia patients. However, the reliability and practicality of such systems remains unclear. In this work, we investigate the potential limitations and challenges in developing a robust MCI detection method using two contrasting datasets: 1) CAUEEG, collected and annotated by expert neurologists in controlled settings and 2) GENEEG, a new dataset collected and annotated in general practice clinics, a setting where routine MCI diagnoses are typically made. We find that training on small datasets, as is done by most previous works, tends to produce high variance models that make overconfident predictions, and are unreliable in practice. Additionally, distribution shifts between datasets make cross-domain generalization challenging. Finally, we show that MCI detection using EEG may suffer from fundamental limitations because of the overlapping nature of feature distributions with control groups. We call for more effort in high-quality data collection in actionable settings (like general practice clinics) to make progress towards this salient goal of non-invasive MCI detection.
△ Less
Submitted 15 January, 2025;
originally announced January 2025.
Interactions Between Solar Neutrinos and Solar Magnetic Fields
Authors:
D. S. Oakley,
Herschel B. Snodgrass
Abstract:
We attempt to correlate all of the available solar-neutrino capture-rate data with the strong magnetic fields these neutrinos encounter in the solar interior along their Earth-bound path. We approximate these fields using the (photospheric, magnetograph-measured) surface magnetic flux from central latitude bands, time delayed to proxy the solar interior. Our strongest evidence for anticorrelatio…
▽ More
We attempt to correlate all of the available solar-neutrino capture-rate data with the strong magnetic fields these neutrinos encounter in the solar interior along their Earth-bound path. We approximate these fields using the (photospheric, magnetograph-measured) surface magnetic flux from central latitude bands, time delayed to proxy the solar interior. Our strongest evidence for anticorrelation is for magnetic fields within the central $\pm 5^{0}$ solar-latitude band that have been delayed by 0.34 years. Assuming a neutrino-magnetic interaction, this might indicate that interior fields travel to the solar surface in this period of time. As more solar-neutrino flux information is gathered, the question of whether such evidence is physical or statistical in nature should be resolved, providing that new data span enough solar cycles and that correlation studies focus on these isolated fields.
△ Less
Submitted 4 April, 1996;
originally announced April 1996.
